1. Field of the Invention
The present invention relates to filtration membranes. More particularly, the invention relates to polymeric ultrafiltration membranes with enhanced flow rates, and to a simplified method of their manufacture.
2. Background of the Technology
Filtration membranes are useful for numerous applications wherein it is desirable to purify or separate components of gaseous or liquid mixtures. Membranes may be classified based in part on these uses. One classification scheme lists types of membranes functionally in increasing order of their size selectivity: gas separation (GS), reverse osmosis (RO), ultrafiltration (UF), and microfiltration (MF).
UF membranes typically are characterized as having a particular molecular weight cutoff or exclusion. For example, a membrane capable of retaining or excluding 90% or more of a macromolecule of 100,000 daltons could be classified as a 100K filter. Likewise, a membrane effective in efficiently excluding or retaining a macromolecule of 10,000 daltons may be referred to as a 10K filter. Of course, a 10K filter not only retains macromolecules having a MW of 10,000 daltons, but also generally retains any larger macromolecules in a heterogeneous solution with equal or greater efficiency. This fact accounts for the problem of membrane clogging that is inversely proportional to the size exclusion rating of the membrane--those membranes with a smaller exclusion limit retain more macromolecules in a heterogeneous mix, and tend to clog or foul more rapidly than membranes with a higher MW exclusion rating.
A particular challenge in the manufacture of UF membranes is to create a membrane with pores that are small enough to achieve an effective retention of macromolecules, while maintaining an acceptable flow rate of the fluid to be filtered. A membrane's resistance to fluid flow is a function of the diameter of the smallest, or retentive, pores through which the fluid must pass, and is also a function of the thickness of the layer of retentive pores. All other factors being equal, membranes whose limiting pores are situated in a relatively thin layer will have faster flow rates than membranes with a thicker layer of limiting pores.
In principle, an optimal membrane would have all of its limiting pores in one layer that is one molecule thick. Clearly, such a membrane would be practically impossible to manufacture or to handle after manufacture. This is especially true in the UF range and below (RO and GS ranges), because the membranes used in these applications often must structurally withstand relatively high pressures that are applied to accelerate the filtration process or to provide energy to overcome diffusion kinetics or osmotic forces. The different structural configurations of UF membranes, namely composite and integrally skinned membranes, represent practical attempts to approach the ideal of a very thin retentive layer.
Composite membranes have a relatively thin layer of retentive pores laminated to a support structure, which is often of a composition that is different from that of the retentive layer. The support structure stabilizes the retentive layer for ease of handling and is intended to offer very little resistance to fluid flux. Composite UF membranes are disclosed in PCT International Publication No. WO 96/02317. In this publication, some disadvantages of composite UF membranes are evident. For example, high filtration pressures can cause delamination of the composite and subsequent membrane failure. Additionally, many composites require at least a two stage manufacturing process to form sequentially the layers of which the composite is made.
The alternative to a composite membrane is an integral membrane. The term integral simply means that the membrane is all of one composition and is prepared in one casting process, although integral membranes may have structurally distinct regions or layers within the integral cross section. Integral UF membranes typically have a retentive skin in contact with a more porous support structure. In such a membrane, the main function of the support structure is to provide thickness to the membrane for ease of handling. In cross section, this kind of membrane displays a relatively dense skin at one surface, with an abrupt transition to a matrix of much larger pores in the support structure.
A disadvantage of this type of membrane, especially in the UF range, is that the support structure has numerous large voids known as macrovoids. These are finger-like projections in the support structure that generally do not communicate with the pores in the skin surface. Fluid entering a macrovoid is trapped and cannot be filtered. Macrovoids in a membrane therefore add to the membrane's resistance to fluid flow, leading to undesirably low flow rates without any concomitant benefit in effectiveness of filtration. Therefore, while a macrovoid-ridden support structure may provide mechanical stability to the skin, this configuration does not result in optimal UF membrane performance.
Skinned UF membranes are disclosed in U.S. Pat. No. 4,481,260 to Nohmi. The '260 patent is directed to hollow fiber membranes for UF applications. Hollow fiber membranes are commonly used for large-scale filtration operations wherein it is desirable to maximize the filter surface area for high total throughput. Typically, hollow fiber membranes must be spun from relatively viscous dope mixes with high concentrations of total solids. The structure of the Nohmi membranes is characterized by a skin layer with pore sizes effective for UF, adjacent to a support structure replete with macrovoids. Accordingly, the performance of these membranes is limited by the presence of macrovoids in the support structure. Also, the formulations of the Nohmi patent and other hollow fiber membrane disclosures are generally not appropriate for or applicable to flat sheet membranes, because of the higher dope viscosity and total solids required for the spinning of hollow fiber membranes as compared to formulations that may be suitable for casting flat sheet membranes.
Because the average pore diameter of the skin layer of a skinned membrane is so different from the average pore diameter of a macrovoid-containing support structure, such membranes have been called asymmetric, or anisotropic. This term is now often broadly applied to any membrane that has cross-sectional gradations in pore diameter, whether the gradations be abrupt or gradual. However, included among membranes that can be called asymmetric are very different structures that represent extremes of performance.
The highly asymmetric membrane structure pioneered by U.S. Pat. Nos. 4,629,563 (Reexamination Certificate No. B1 4,629,563, issued Jun. 3, 1997) and 4,774,039 to Wrasidlo, represents a vast improvement over the so-called asymmetric membranes that preceded it. This is because, in the MF range, the Wrasidlo membrane structure is free of macrovoids. Without macrovoids, the dead space within the membrane is significantly reduced if not eliminated, and flux rates are improved over prior MF membranes. In addition, the high degree of asymmetry within the support structure is gradual, rather than abrupt. This allows the support structure to act as a prefilter (or more accurately, as several prefilters of different sizes), and greatly enhances the life and dirt-holding capacity of the membrane by retaining particles that are much larger than the skin pores well before they enter the region of the skin layer.
The manufacture of the Wrasidlo MF membranes is based on the properties of an unstable dispersion of a membrane dope mix within the binodal or spinodal curves of a phase diagram. In this method, care must be taken with each casting to assure that the unstable dispersion is within the binodal or spinodal in order to achieve the desired membrane structure. The dope mix is typically under constant agitation prior to casting to prevent premature phase separation. Despite the prophetic UF membrane examples in the Wrasidlo patents, our diligent efforts to produce highly asymmetric, macrovoid-free UF membranes from an unstable solution have been unsuccessful. In membranes having skin porosities appropriate for UF applications, there have always been significant macrovoids present in the support structure, similar to the structures disclosed in the Nohmi patent.
Accordingly, it would be highly desirable to prepare UF membranes having a highly asymmetric structure analogous to the Wrasidlo membranes. Further, it would be beneficial to create such membranes by a method that does not require the careful and constant attenuation of a casting dope in an unstable dispersion. Such an advance would produce UF membranes with enhanced flow rates and dirt-holding capacity, and the manufacture of these membranes would be greatly simplified.